Re: Off topic - personal problem in VT

Fred Weston wrote:
> Jim Thompson wrote:
> 
>> 100m (328') is the "textbook" max length of Ethernet, but this has a 
>> lot more to do with signaling collisions than the ability to correctly 
>> drive signals on the wire.
>>
>> Put a switch (or full-duplex adapter) on each end, and "collisions" 
>> become impossible.
> 
> 
> That's interesting.  So you're basically saying that if you eliminate 
> the possibility of collisions on the wire, you can run ethernet much 
> further than 100 meters?  

The collision domain is where the 100m "limit" comes from.

The smallest Ethernet frame is 64 bytes, or 512 bits.
A 10Mbps system running over copper wire will carry bits at 0.75c, or 
224,844 kilometers per second. This means that 256bits at 10Mbps will be 
contained in 5,756 m of copper cable. The original DIX Ethernet design 
specifications allowed for a total of three 500m runs of copper cable, 
plus allowance for 2 repeaters, and a generous allowance for error.

Getting 100Mbps Ethernet to run over twisted pair beyond 100m is going 
to be much tougher than it will be @ 10Mbps.

 > Do you have any idea what the max
> "transmission" length is?

It depends on the PHYs at both ends, and how good the wire is.  AMD sold 
"extended distance" PHYs for a while.  Someone else did too, and I 
forget the manufacturer.  When I was engineering "Everywire" for 
Wayport, I speced extended reach PHYs for the four uplink ports out of 
each unit, just so we could cross a hotel if needed.   I saw those run 
150m on  cat5.  Its unlikely that you can find these anymore.

I've seen low-attenuation cable marketed, with claims for 160m.  I have 
no experience with this though.

Jim
_______________________________________________
Talk mailing list
[email protected]
http://seattlewireless.net/mailman/listinfo/talk